UBC Faculty Research and Publications

A search for protein biomarkers links olfactory signal transduction to social immunity Guarna, Maria M; Melathopoulos, Andony P; Huxter, Elizabeth; Iovinella, Immacolata; Parker, Robert; Stoynov, Nikolay; Tam, Amy; Moon, Kyung-Mee; Chan, Queenie W; Pelosi, Paolo; White, Rick; Pernal, Stephen F; Foster, Leonard J Feb 8, 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12864_2014_Article_1193.pdf [ 2.49MB ]
JSON: 52383-1.0223384.json
JSON-LD: 52383-1.0223384-ld.json
RDF/XML (Pretty): 52383-1.0223384-rdf.xml
RDF/JSON: 52383-1.0223384-rdf.json
Turtle: 52383-1.0223384-turtle.txt
N-Triples: 52383-1.0223384-rdf-ntriples.txt
Original Record: 52383-1.0223384-source.json
Full Text

Full Text

RESEARCH ARTICLEA search for protein biomcetiecagriculture exceeds $2.3 billion (Alex Campbell, Agriculture creased to an average of approximately 30% [2]. WhileGuarna et al. BMC Genomics  (2015) 16:63 DOI 10.1186/s12864-014-1193-6problem [4]. V. destructor is now considered the singleVancouver, BC V6T 1Z4, CanadaFull list of author information is available at the end of the article& Agri-Food Canada. Personal communication) while thevalue added to crops such as almonds, berries, fruits,vegetables and other nuts in the U.S. is estimated tobe $11.7 billion [1]. Winters, in particular, are a pro-found determinant of colony survival. Prior to 2006,this has not yet had a discernable effect on crop yields,it has made it much more challenging for beekeepingcompanies to remain solvent. The causes of honey beelosses have been attributed to a multitude of factors[3], including bee-specific pathogens and parasites such asthe mite Varroa destructor and the microsporidia Nosemaapis and Nosema ceranae. Also, long-implicated as a lead-ing cause of colony mortality before the introduction of V.destructor, the bacterial brood pathogen Paenibacillus lar-vae that causes American Foulbrood continues to be a* Correspondence: marta.guarna@outlook.com; foster@chibi.ubc.ca1Department of Biochemistry & Molecular Biology, Centre forHigh-Throughput Biology, University of British Columbia, 2125 East Mall,of crops. The contribution of honey bees to Canadianpollination activities. For years, beekeepers have controlled deadly pathogens such as Paenibacillus larvae, Nosema spp.and Varroa destructor with antibiotics and pesticides but widespread chemical resistance is appearing and mostbeekeepers would prefer to eliminate or reduce the use of in-hive chemicals. While such treatments are likely to still beneeded, an alternate management strategy is to identify and select bees with heritable traits that allow them to resistmites and diseases. Breeding such bees is difficult as the tests involved to identify disease-resistance are complicated,time-consuming, expensive and can misidentify desirable genotypes. Additionally, we do not yet fully understand themechanisms behind social immunity. Here we have set out to discover the molecular mechanism behind hygienicbehavior (HB), a trait known to confer disease resistance in bees.Results: After confirming that HB could be selectively bred for, we correlated measurements of this behavior withprotein expression over a period of three years, at two geographically distinct sites, using several hundred bee colonies.By correlating the expression patterns of individual proteins with HB scores, we identified seven putative biomarkers ofHB that survived stringent control for multiple hypothesis testing. Intriguingly, these proteins were all involved insemiochemical sensing (odorant binding proteins), nerve signal transmission or signal decay, indicative of the series ofevents required to respond to an olfactory signal from dead or diseased larvae. We then used recombinant versions oftwo odorant-binding proteins to identify the classes of ligands that these proteins might be helping bees detect.Conclusions: Our data suggest that neurosensory detection of odors emitted by dead or diseased larvae is the likelymechanism behind a complex and important social immunity behavior that allows bees to co-exist with pathogens.BackgroundThe health of honey bees (Apis mellifera L.) is crucialfor honey production and pollination of a wide varietyoverwintering mortality of honey bee colonies inNorth America was 10 to 15%; however, losses inNorth America and Europe have dramatically in-signal transduction to soMaria Marta Guarna1*, Andony P Melathopoulos2,6, ElizabNikolay Stoynov1, Amy Tam1, Kyung-Mee Moon1, QueenStephen F Pernal2 and Leonard J Foster1*AbstractBackground: The Western honey bee (Apis mellifera L.) is a© 2015 Guarna et al.; licensee Biomed CentralCommons Attribution License (http://creativecreproduction in any medium, provided the orDedication waiver (http://creativecommons.orunless otherwise stated.Open Accessarkers links olfactoryial immunityh Huxter3, Immacolata Iovinella4, Robert Parker1,7,WT Chan1, Paolo Pelosi4, Rick White5,ritical component of human agriculture through its. This is an Open Access article distributed under the terms of the Creativeommons.org/licenses/by/4.0), which permits unrestricted use, distribution, andiginal work is properly credited. The Creative Commons Public Domaing/publicdomain/zero/1.0/) applies to the data made available in this article,profiles could be accurately characterized then the pro-Guarna et al. BMC Genomics  (2015) 16:63 Page 2 of 16greatest natural threat to honey bees worldwide, as itweakens and kills colonies by parasitizing bees as well asvectoring several viruses that may be even more virulentto bees than the mites themselves [5].Though acaricides, antibiotics and fungicides are regis-tered for controlling V. destructor, P. larvae and Nosemaspp., a number of negative consequences are associatedwith their use. These include the economic cost of thetreatments themselves, concerns around the potentialcontamination of hive products [6], widespread anti-biotic [7] and acaricide resistance [8-11], and concernsover the effectiveness of chemotherapy for controllingNosema spp [12]. Indeed, these pathogens are on a pathakin to a chemical treadmill whereby resistance developswithin a few years of the initial use of a particular chem-ical [13]. At the same time, viruses’ impact on bee healthcontinues to increase and no conventional treatmentsare available to counter them. The phenomenon of drugresistance is not recent, even in microlivestock such asbees, and will likely become more widely spread. Spivakand Gilliam observed over ten years ago that acaricidesand antibiotics were no longer effective against Varroaand Paenibacillus larvae [14].An alternative pest management approach is to iden-tify and select bees with an increased ability to toleratediseases without chemical intervention. While this couldbe achieved through heightened innate immunity, honeybees, as eusocial animals, have the added capacity for so-cial behaviors that enable resistance to pathogens. Sev-eral behaviors that help to confer colony-level resistanceagainst parasites and pathogens have been characterized,including hygienic behavior (HB) [15,16], Varroa Sensi-tive Hygiene (VSH) [17], grooming behavior [18,19] andothers [14]. HB is the best understood and it involvesthe detection of dead or diseased bees in brood cells, un-capping of cells and removal of the affected larvae orpupae by nurse bees. The primary means by which HBconfers disease resistance is thought to be the continualelimination of brood pathogens from the hive environ-ment, which would otherwise remain, multiply and po-tentially infect other bees. In the case of P. larvae, withwhich A. mellifera has had the longest time to co-evolve,bees specifically remove infected larvae or pupae whenthe bacteria are still in the vegetative state [20]. Al-though the term ‘hygienic behavior’ was originally usedto describe removal of brood infected with P. larvae, itsuse has gradually been expanded to describe the removalof brood infected with chalkbrood disease (caused byAscosphaera apis), as well as brood parasitized by Varroa[21]. It has further been applied to brood invaded by thegreater wax moth, Galleria mellonella [22], or broodinfested with the small hive beetle, Aethina tumida [23].The magnitude of HB varies between individual coloniesand populations, making it possible to selectively breedteins most highly up or down-regulated relative to ‘nor-mal’ bees would make excellent biomarkers for selectivebreeding and they would also be likely to have a mech-anistic role in the manifestation of the behavior. HB isheritable so it follows that the expression patterns ofproteins involved in HB must also be heritable so wefirst established that we could selectively breed for HBefficiently.Prior to the experiments aimed to evaluate protein ex-pression patterns and to identify HB markers, we per-formed a small selective breeding program to confirmthat social immunity traits, particularly HB, could beenriched in our apiaries based on field testing. Using aclosed mating system and the field-based liquid nitrogenfreeze-kill test [14] as a test of HB, we observed a signifi-cant enrichment of the trait over only three generations(Figure 1)(P = .0002, F = 8.719, df = 2, one-way ANOVAof frequency distribution data). The distribution of HBin the population changed dramatically in the first roundof selective breeding, and the HB values continued toimprove in the second round of selective breeding. Fur-bees to enrich HB [20,24], with the goal of increasing theability of the bee population to manage disease while re-ducing beekeeper intervention and chemical treatments.Testing colonies for selective breeding, however, is a slowand resource-intensive process that could be facilitated bymore rapid molecular assays.We have previously reported that larval proteins in-volved in chitin biosynthesis, wound healing and innateimmunity pathways are correlated with HB and VSH[25]. In the same study, we discovered that HB andVSH were also associated with the expression of anten-nal proteins in functional classes such as cell surface-linked signaling and protein modification pathways. Inthe current study, we extend our original approach tocover three generations, two geographically distinctsites and far more colonies (N = 167) towards two goals:1) identify prognostic biomarkers that could be usedfor marker-assisted selection and 2) better understandthe mechanisms underlying behaviors that confer socialimmunity.ResultsEnrichment of HB by selective breeding in GFDifferences in a phenotype such as a behavior must bethe result of altered protein expression or activity so ourguiding hypothesis was that honey bees exhibiting hy-gienic and other social immunity behaviors should dis-play unique protein expression profiles, at least in thetissues/organs involved in the behaviors. If these uniquether, the proportion of colonies with high HB scores(higher than 90% cells removed after 24 h) increasedGuarna et al. BMC Genomics  (2015) 16:63 Page 3 of 16Figure 1 Selective Breeding of hygienic behavior (HB). In an initialbreeding experiment, colonies were selectively bred for HB over threegenerations to test the efficacy of selection. For each generation,colonies expressing the highest level of HB estimated by the fractionof freeze-killed pupae removed after 24 hours (R24) were selected asbreeders and their virgin queens and drones close mated in an isolatedapiary. F0 bars represent the natural frequency distribution of HB in thestarting population, while the F1 and F2 bars represent the distributionfrom 12% in the first year to 18% the second year and to26% in the third year.Wide range of hygienic behavior in starting populationsWe then explored the correlation between protein ex-pression and HB to identify proteins correlated with thebehavior. We established two populations in geographic-ally distinct areas of Western Canada and proceeded tobreed progeny with divergent levels of HB and measuredprotein expression profiles for each generation (Figure 2).The starting colonies at each location presented a widerange of HB (Figure 3) and were assembled from geo-graphically diverse sources of stock used by Canadianbeekeepers. Breeding was performed as outlined inMethods using instrumental insemination of drones andvirgin queens from high- and low-scoring colonies wasused to achieve a controlled partial diallel crosses whichcreated high and low scoring colonies, as well as hybrids,intended to facilitate the identification protein expres-sion related to HB, and to estimate heritability of theprotein markers (Figure 2).Antennae of workers on brood frames were collectedfrom all colonies in each of the generations. The choiceof tissue to focus on is obviously important when study-ing expression markers (as opposed to quantitative traitafter one and two rounds of selection, respectively. The frequencydistribution in the F1 and F2 is significantly different from the F0(*P = .0002, one-way ANOVA).loci where all cells and tissues would be expected tohave the same genotype) and as such we were guided bythe following rationale: what causes bees to be hygienicis not known but it may involve either a heightened sen-sitivity to detect a specific signal or a unique wiring ofthe brain that causes hygienic bees to respond to a signaldifferently. The former possibility would likely manifestas differences in sensory organs (antennae are bees’ pri-mary sensory organ) while the latter would mean differ-ences in their brains. Le Conte et al. [26] reported theanalysis of brain transcriptome of highly Varroa-hygienicbees and the identification of a set of genes involved insocial immunity. Although the function of these candi-date genes did not seem to support higher olfactory sen-sitivity in hygienic bees as previously hypothesized, theauthors noted that the analysis of peripheral tissues, likeantennae, should be performed since insect behavior canbe dramatically affected by changes in expression ofantennal-specific genes. We, therefore, elected to focuson the proteomic analysis of antennae. Proteins were ex-tracted from the collected antennae and the protein ex-pression pattern of each colony was measured by massspectrometry, employing a triplexed stable isotope label-ling approach that we have applied previously [25] in arandomized incomplete block (Figure 4a, see Methods).Approximately 1300 proteins were quantifiable in theanalyses of each generation and, of these, approximately500 in each set were represented in at least 25% of the‘blocks’ (triplex-labelled samples, Figure 4b). Those de-tected in at least 25% of the blocks were then comparedwith HB data collected for the same colonies in order toidentify proteins whose expression patterns correlatedwith the behavior (see Methods). Briefly, a Linear MixedEffects model was used to estimate the effect of eachpredictor variable (e.g. geographic origin of the popula-tion and HB) on the protein expression level.Geographical origin affects protein expressionGiven our previous observations [27] we first askedwhether any of the protein expression profiles could beexplained, at least in part, by the geographical origin ofthe colonies? Of the 476 proteins in at least 25% of theblocks, the profiles for thirty eight showed a statisticallysignificant effect of geographical origin (Q < .05) andthese were then subjected to gene enrichment analysisas described previously [27]. Hierarchical clustering ofheat-maps enabled the visualization of two major proteinclusters associated with specific biochemical functions(Figure 5, Additional file 1: Table S2). Enrichment ana-lysis indicated that bees of Canadian and Californian ori-gin are divergent in expression of proteins involved inmitochondrial respiration (e.g. ATP-synthase, CytochromeC) and glutathione detoxification (Glutathione-S transfer-ase S1, Peridoxin). The Chilean bees were divergent,Figure 3 Hygienic behavior of starting populations. A. HB of the Y1 colonies in Grand Forks, BC and at the Research Farm in Beaverlodge,Alberta, showed a wide range of HB values as determined by the proportion of cells that were uncapped and pupae were removed after 24 h(R24h) using the liquid nitrogen freeze-killed brood method. B. The five populations analysed in Beaverlodge, one originated in Saskatchewan(SK), two in California (CA1 and CA2), one in Chile (Ch) and one in Ontario (ON), also showed a wide range of HB.Figure 2 Breeding and sampling design for the identification of heritable protein markers. A. Diagram illustrating our approach to identifyspecific proteins associated with HB behavior using mass spectrometry. B. Breeding design and origin of samples. Parallel breeding programswere conducted in Beaverlodge, Alberta (55°N, 119°W) and Grand Forks, British Columbia (49°N, 118°W). Samples for proteomic analysis weretaken each year at each location. The partial diallel cross used with the instrumental insemination in AB and in Y3 in BC is shown in C.Guarna et al. BMC Genomics  (2015) 16:63 Page 4 of 16sharing reduced expression of mitochondrial enzymeswith Californian and increased expression of Glutathionedetoxification proteins with Canadian varieties.The effect of population origin on the relative abun-dance of proteins in the antenna supports our previousobservation that historic latitudinal clines are detectablein the expression of metabolic proteins in imported bees[27]. The co-regulation of proteins driving oxidative me-tabolism with those of glutathione mediated detoxifica-tion is consistent with the management of oxidativeFigure 4 Structure and results of the randomized incomplete block design. A. The matrix illustrated below allowed for a comparison of theprotein profiles of all colonies. Each colony was sampled in triplicate and each sample was labeled with either light, medium or heavy isotopes,and was assigned to a block. Each block, represented by a triangle, contained three samples from different colonies and labels and was analysedin one Mass Spectrometry run. B. Due to the semi-stochastic nature of data-dependent acquisition in LC-MS/MS and the wide range of proteinabundances, not all proteins were detected in every sample. Shown is a representative plot of the frequency where 1312 proteins were detectedacross 38 blocks in BL-Y1 sample set. The 476 proteins that were detected in at least 10 (25%) of the blocks were considered in the analysis.Guarna et al. BMC Genomics  (2015) 16:63 Page 5 of 16Figure 5 Clustering and functional enrichment of proteins regulated byclustered (SOTA) based upon statistical significance and relative abundance. Uproteins were predicted and each cluster was tested for the enrichment of gemolecular function using DAVID (see Methods).geographical origin. Proteins regulated by geographical origin weresing orthologous proteins derived from Drosophila the function of beene ontology categories biological processes, cellular compartment andstress associated with increased metabolic rates. In ouranalysis of population origin, no functional enrichmentfor neurological proteins was observed compared to theantenna proteome indicating that basic neurologicalfunction or structure was similar between the bees ofdifferent geographical origin.Proteins correlated with HBWe then studied whether variations in protein expres-sion were associated with behavioral differences, and ifthey could be predictive markers of social immunity, fo-cusing on HB (Additional file 2: Table S3). Among theY1 colonies at the Alberta breeding site, the expressionpatterns of GI:48138819 (XP_393426.1) and GI: 66512788(XP_392466.2) showed a strikingly strong correlation withHB (P < .0005, Q < .1, Figure 6). Proteins that are intrin-sically involved with a behavior should be correlatedwith that behavior independent of where the measure-ments are taken and as such we examined whichproteins were consistently correlated with HB at bothbreeding sites. The proteins strongly correlated with HB(P < .05) at both locations were Odorant binding pro-tein 16/GI:94158709, Calcyclin Binding Protein/GI:66564402, 3-ketoacyl-CoA thiolase/GI: 48097100, andVAMP (vesicle-associated membrane protein)/GI:48138819,found previously. The other protein observed initially, GI:66512788, had too many missing data at the GF breedingsite to be considered. We then extended this approach tocover Y1, Y2 and Y3, particularly to incorporate heritabilityinformation obtained from the Y3 BC partial diallel cross(Figure 2). Over the six datasets (Figure 2), proteins wereranked based on a HB-correlation factor average of all data-sets plus a heritability factor (see Methods). Of the topranked proteins, we also required that they were quan-titated in at least four of the six datasets. BM-40-SPARC, Odorant binding protein 18/GI: 110774625 and26S protease regulatory subunit 6A/GI: 48101907 showedthe highest HB correlation factors when considering allionI: 6Guarna et al. BMC Genomics  (2015) 16:63 Page 6 of 16Figure 6 Identification of proteins correlating with HB. A. Partial regressproteins with a strong correlation with HB. The P-values for BM-40-SPARC/ Gand the Q-values were 0.05562 for both proteins. B. Final HB correlation score oSPARC, Odorant binding protein 18/ GI:110774625 and 26S protease regulatoryplots of HB versus protein level. Results from BL-Y1 colonies showed two6512788 and VAMP/ GI:48138819 were 0.00017 and 0.00024, respectivelyf proteins from all datasets calculated as described in Methods. BM-40-subunit 6A/ GI:48101907 showed an HB correlation factor >5.datasets. Compiling these approaches, the seven proteinsthat emerged (Table 1) with expression patterns weremost closely correlated with HB, are thus putatively usefulfor marker assisted selective breeding of HB and also sug-gest some testable hypotheses regarding the mechanism(s)underlying HB.Exploring potential ligands of OBPs linked to HBThe three OBPs bind N-phenyl-1-naphthylamine (1-NPN) with dissociation constants in the range of 1 to5 μM (Figure 8), similar to most other insect OBPs[29,32], thus allowing accurate evaluation of the affinitiestowards other ligands in competitive binding assays. AllIC50 and Kd values are listed in Additional file 3: TableS4, while the dissociation curves for the strongest li-gands are shown in Figure 7. In general, all three OBPstended to prefer fatty acids and their ester and amide de-Table 1 Proteins most highly correlated with HBAccession # Protein markerse CGuarna et al. BMC Genomics  (2015) 16:63 Page 7 of 16GI: 110774625 Odorant binding protein 18GI: 94158709 Odorant binding protein 16GI: 48138819 VAMP (vesicle-associated membrane protein)GI: 66512788 BM-40-SPARC (Secreted protein acidic and rich in cysteinGI: 66564402 Calcyclin Binding ProteinAmong the seven HB-associated proteins (Table 1) twowere odorant-binding proteins (OBPs), suggesting thathygienic bees may be sensing something emitted fromdiseased or dying larvae. The natural ligands are un-known for most bee OBPs, however, so we purified re-combinant versions of the two high correlated OBPsdetected here (OBP16, OBP18) and screened their af-finity towards several metabolites in competitive bind-ing assays. As reference, we chose OBP21, which is a‘C-minus’ OBP, like OBP16 and OBP18, containing onlyfour of the six conserved cysteines present in classicOBPs [28]. We had also characterized OBP21 in termsof ligand-binding affinities [29].Affinities of twenty-nine ligands to the three proteinswere evaluated in competitive binding assays, using N-phenyl-naphthylamine (1-NPN) as a fluorescent reporterand measuring the fluorescence decrease produced by theaddition of the ligand in a concentration dependent fash-ion. Actual dissociation constants were calculated fromthe concentrations of each ligand halving the initial fluor-escence of the complex ([IC]50) as described in Methods.Potential ligands were selected from among the terpe-noids commonly occurring in the scent of flowers, fattyacids and their esters, components of the queen man-dibular pheromone and the brood pheromone andother volatiles potentially linked to hygienic behavior[30,31]. Figure 7 shows the binding affinity of thetwenty-nine compounds tested; the upper panel reportsthe data for the strongest ligands, whose affinities areabout one order of magnitude higher than the ligandsreported in the lower panel. Interestingly, OBP16 andOBP18 bind to nearly all compounds tested with higheraffinity than OBP21.GI: 48097100 3-ketoacyl-CoA thiolase, mitochondrial-likeGI: 48101907 26S protease regulatory subunit 6Arivatives, as well as three structurally similar terpenoids:3,7-dimethyloctanol, geraniol and geranyl acetate. OBP16and OBP18 can both accept linear and branched mole-cules while OBP21 seems to have greater specificity to lin-ear chains of 18 carbon atoms. OBP18 showed the highestaffinities for several compounds including oleic acid whichis released by decaying insect corpses [33].DiscussionHoney bees are an essential component of human agri-culture but are under ever-increasing threat from para-sites and infectious diseases. While acaricides, fungicidesand antibiotics have been useful for controlling manybee pathogens and pests, resistance to these products isspreading and there is substantial public pressure tomove away from such chemical controls. In-hive chem-ical treatments can leave residues in honey and otherhive products, and in the environment. In addition, theycan have sublethal effects on bees including potential ef-fects on their immune system [34]. Selection of beeswith higher disease resistance may reduce the need ofchemical treatments and some disease-resistance traitshave been identified in bees. Though honey bees can beselectively bred, the traits for disease resistance are,however, difficult to test and only a few, highly special-ized groups, usually at universities or in technical trans-fer teams are able to use them (e.g., hygienic behavior,grooming, Varroa-sensitive hygiene) [35]. If molecularmarkers correlated with disease resistant traits wereknown, bee breeders could potentially utilize these in amore effective selective breeding program by employingmolecular diagnostics in place of field behavioral assays.One typically thinks of such markers at the level ofDirection General function↑ Ligand binding↑ Ligand binding↓ Nerve signal transductiona binding) ↑ Nerve signal transduction↑ Signal down-regulation↑ Ligand degradation↑ Signal down-regulation via protein degradationGuarna et al. BMC Genomics  (2015) 16:63 Page 8 of 16DNA (e.g., quantitative trait loci, single-nucleotide poly-morphisms) but there has been little progress towardsidentifying any loci linked to disease-resistant behaviorsin bees [36,37]. Nevertheless, protein expression levelscould be marker of disease resistance. Indeed, the pro-teins identified would likely be more closely linked tothe mechanism behind the behavior than a DNA feature.Protein biomarkers and prospects for marker-assistedselective breeding in beesHB is an economically beneficial, heritable trait that en-ables bees to co-exist with pathogens and as such wehave undertaken an extensive and exhaustive search forproteins whose expression levels are highly correlatedwith HB. We chose to focus on expression levels withinthe antennae of nurse bees because it is this particularFigure 7 Potential ligands of OBPs. Twenty nine potential ligands were tesand OBP21. Each purified recombinant protein was mixed with the fluorescen2 μM in Tris buffer. Each mixture was titrated with a ligand and the displacemrelative dissociation constants, as described in the Methods section. Compounwith lower affinity are shown in the lower panel. OBP16 and OBP18 showed hOBP21. Of particular interest is the high binding affinity of OBP18 to oleic acidbehavioral ontogenic stage that performs the HB in thecolony and one of the likely mechanisms for HB is aheightened ability to sense dead, dying or diseased larvae[38], which would likely involve antennae. Despite theseexperiments being performed in the field with geneticallydiverse, outbred populations, the results provided a strik-ingly clear confirmation of our hypothesis, that there is in-deed a protein expression pattern unique to HB.The discovery of biomarkers specific to HB allows forthe possibility to develop prognostic assays that could beused to select parent colonies in a marker-assisted selec-tion breeding program. We envision that the HB markersreported here will be useful to facilitate selective breedingefforts. Future work will aim to validate and apply thesebiomarkers in a marker-assisted selective breeding pro-gram with the goal of enriching honey bee populations forted in a competitive binding assay for their binding to OBP16, OBP18t probe N-phenyl-1-naphthylamine (1-NPN), both at the concentration ofent of the fluorescent probe from the complex was used to evaluate theds with higher affinities are shown in the top panel, while compoundigher affinities for most compounds than the previously characterized.Guarna et al. BMC Genomics  (2015) 16:63 Page 9 of 16ACthis social immunity trait. Beyond this potentially practicalapplication, the proteins whose expression levels weremost highly correlated with HB are all obviously linked tovarious aspects of chemosensory processes, suggestingseveral very interesting and testable hypotheses regardingthe mechanism(s) underlying HB.Sensing the signal for HBInsects rely on chemical communication to monitor theenvironment and exchange information between conspe-cifics. Social insects, in particular have developed ahighly sophisticated chemical language, enabling mem-bers of the colony to perform different tasks. Thechemoreception system of insects is mediated by olfactoryreceptors, located on the membrane of sensory neurons,and by soluble proteins present at high concentration inthe lymph of chemosensilla [32,39]. These proteins belongto two major classes, OBPs (odorant-binding proteins)and CSPs (chemosensory proteins), in both cases smallpolypeptides folded into α-helical domains, arranged intwo different unique motifs [40]. Although the specific ac-tion of OBPs and CSPs is not yet clear, certainly airbornemolecules, as those associated with diseased or dead bees,upon entering the antenna interact with these solubleFigure 8 Differential ligand binding to OBPs. A. Binding curve of OBPs to 1allowed for an accurate evaluation of the affinities towards other ligands in comof 2 μM in Tris buffer were titrated with increasing amounts of the fluorescent phighest affinities for OBP16, OBP18 and OBP21 respectively using the procedureBDproteins and are tightly bound to be carried through theaqueous environment of the sensillar lymph to membranereceptors [39]. Thus, if hygiene in bees is due to an en-hanced sensitivity towards specific signals originated inthe affected brood, OBPs would probably be effecting thisand OBP16 (GI: 94158709) and OBP18 (GI: 110774625)are the most likely candidates, based on the data pre-sented here. Neither OBP16 nor OBP18 are exclusive toantennae, although their tissue expression patterns areconsistent with a sensory molecule. OBP16 is expressedexclusively in peripheral tissues, including antennae, andis found in all castes, although it is most highly expressedin workers [29,41,42]. OBP18 is also largely in peripheraltissues but is also concentrated in the nerve cord of the fe-male castes [41]. Clearly this is where OBPs should be, butwhat is it that they are detecting when they are expressed?The natural ligands of most bee OBPs, including thesetwo, are unknown but our investigations here with recom-binant proteins suggest that both OBP16 and OBP18 pre-fer branched and linear fatty acids. This general class ofmolecules includes many bee pheromones [43] so is con-sistent with a role for them in HB. Of particular interest isthe high affinity of OBP18 to oleic acid because it is re-leased by decomposing insects [33] and may be a strong-NPN, showing dissociation constants in the range of 1 to 5 μM whichpetitive binding assays. Purified recombinant proteins at the concentrationrobe. B, C and D: Competition curves for the seven ligands with thedescribed in Figure 7 and in the Methods section.Guarna et al. BMC Genomics  (2015) 16:63 Page 10 of 16mediator for social immunity in Apis mellifera and othereusocial species. Further demonstration of the precise lig-and(s) they are detecting would require electroantenno-gram tests of selected compounds in bees that have hadone OBP or the other knocked down by siRNA.Transmitting the signal for HBTwo proteins involved in inter-nerve communicationwere the most tightly linked to HB: vesicle-associatedmembrane protein (VAMP, GI: 48138819) and secretedprotein acidic and rich in cysteine Ca binding (BM-40-SPARC, GI: 66512788). As any animal behavior requiresperipheral, if not also central nervous system activity, itseems reasonable that a heightened behavior could resultfrom up- or down-regulation of proteins required in sig-nal propagation. VAMP is a well-known SNARE proteinrequired for fusing synaptic vesicles at the synaptic cleftto release neurotransmitters [44] and as such our obser-vation that it is inversely correlated with HB would sug-gest that it is particularly important in an inhibitorysynapse and that its expression needs to be suppressedfor neurons involved in HB to fire properly. BM-40-SPARC (a.k.a. testican in mammals) is a proteoglycanwhose function is not understood but its transcript isup-regulated in nurse bees [45], which are the bees thatperforms HB, and it is implicated in brain developmentin mammals [46].Degrading the signal for HB and down-regulating theresponseAn important aspect of a response to any signal, particu-larly the depolarization involved in triggering a nerve re-sponse, is the termination of the signal to allow thesystem to be reset so that it may respond again. Theremaining three proteins correlated with HB appear tofall within this this category:3-ketoacyl-CoA thiolase (EC2.3.1.16, GI: 48097100) isinvolved in beta-oxidation and catalyzes the conversionof acyl-CoA and acetyl-CoA to CoA by itself and 3-oxoacyl-CoA. The isoform found here is likely the mito-chondrial version, which would suggest that its key roleis in energy production. This could indicate a specificenergy requirement for hygienic behavior but this en-zyme also happens to degrade the same class of mole-cules that acts as ligands for OBP16 and OBP18 so it istantalizing to speculate that it may also act to shut downthe signal for HB.Calcyclin binding protein (GI: 66564402) is involved intargeting specific signalling proteins for degradation inother organisms, implying that it may be involved in de-grading components of the signalling involved in HB,perhaps the OBPs themselves. Calcyclin binding proteinis known as Siah-interacting protein in humans but it isnot clear that its interaction with calcyclin there hasany functional relevance. Structural analysis of Siah andSiah-interacting/calcyclin-binding protein indicates thatcalcyclin-binding protein is a component of an E3 lig-ase complex and that it is required to recruit an E2-substrate complex for the final step of ubiquitin transfer[47]. It has been most-studied in the context of signallingoncogenes so it is attractive to speculate that it may alsobe involved in degrading proteins involved in the signal-ling behind HB and thereby down-regulating the signal.26S protease regulatory subunit 6A (GI: 48101907) ap-pears to be a multi-functional protein and it is not clearwhich of its roles might be relevant in HB. It is a com-ponent of the 26S proteasome [48], which degrades ubi-quitylated proteins, therefore Tat-binding protein mightsomehow help to turn over other proteins directly in-volved in HB, such as those above. Given that calcyclin-binding protein is also involved in ubiquitin-mediatedprotein degradation, it seems most likely that it is in thiscapacity that Tat-binding protein is involved in HB too.However, it is also a transcriptional co-activator [49] ofhormone receptors (a mammalian functional equivalentof OBPs) so it could conceivably be regulating the ex-pression of other proteins involved in HB, such asOBP16 and OBP18.ConclusionsSince protein expression can be influenced by environ-ment and the technology for measuring proteins haslagged behind tools for measuring nucleic acids, proteinmarkers has often been ignored in favor of QTLs orSNPs for marker-assisted selection for breeding pur-poses. Nevertheless, the link between QTLs or SNPs andphenotype can also be influenced by environment inmost cases and so there is no intrinsic reason for pro-teins not to be investigated as biomarkers. To this end,we have shown that the expression levels of a selectedset of proteins are heritably associated with an importantsocial immunity trait in honey bees, hygienic behavior.Our data suggest that bees expressing this trait are bet-ter able to detect and respond to a chemical signal emit-ted by diseased or dying larvae, stimulating them toremove the potential threat from the colony environ-ment. The chemical signal responsible for this remainsto be identified but the proteins described here shouldmake suitable biomarkers to guide selective breeding forhygienic behavior.MethodsEstablishing bee populations, HB testing, breeding andsample collectionThe collection of honey bee samples, field testing andbreeding was performed at two breeding locations inWestern Canada, one near Grand Forks, BC (49°N,118°W), the other at the Research Farm of AgricultureGuarna et al. BMC Genomics  (2015) 16:63 Page 11 of 16and Agri-Food Canada in Beaverlodge, AB (55°N, 119°W).An initial experiment was performed in BC as part of theBC Bee Breeders Association Queen Testing Project toconfirm that hygienic behavior could be selectively bredfor in our apiaries. In this experiment, selection to enrichfor HB was based on field testing using the freeze-killedbrood assay explained below. A second experiment wasperformed both in BC and AB with the aim to correlateproteome profiles with field test results in search for bio-markers of HB. For this breeding and proteomic experi-ment, the year 1 (Y1) colonies in BC included stockspanning a range of HB and Varroa resistance, includinglocal and broader Canadian stock selected for mite resist-ance or HB, as well as descendants of a close-mated popu-lation at the University of Minnesota inbred for hygienicbehavior [20,50] and VSH lines [51,52]. The starting col-onies in AB consisted of eight populations described pre-viously [27] of which five were sampled for proteomicanalysis; these originated from Ontario (ON), California 1(CA1), California 2 (CA2), Chile (Ch), and Saskatchewan(SK). Instrumental insemination (ii) was used for allbreeding, except for the Y2 breeding in BC. Instrumentalinsemination of virgin queens from high- and low-scoringcolonies followed a partial diallel cross design [53] whichcreated high and low scoring colonies, as well as hybrids,intended to facilitate the identification protein expressionpatterns unique to HB. In Y3, we also performed ii of vir-gin queens in BC to evaluate the heritability of the proteinmarkers identified. All inseminated or closed matedqueens were introduced into new colonies. After thequeens started laying, colonies were allowed to developfor at least six weeks to allow worker populations to turnover before they were tested for HB and antennae werecollected for proteomic analysis. Colonies were assessedfor HB using the freeze-killed brood method [14], wherethe proportion of sealed cells that nurse bees uncap (un-capped, U) and remove dead pupae from (removed, R) iscounted at 24 and 48 h using two separate tests performedone week apart on each colony. For proteomic analysis,antennae were cut from adult workers sampled frombrood frames (three pools of ten bees from each colony).Invertebrate research (except on cephalopods) does notrequire ethics certification at our institution.ReagentsAll chemicals used were of analytical grade or betterand all solvents were of HPLC-grade or better; all, withthe exceptions specified below, were obtained fromThermoFisher-Scientific (St. Waltham, MA, USA). Che-micals for protein expression and purification and forbinding assays were purchased from Sigma-Aldrich andwere of reagent grade, with the exception of selectedcompounds used in binding assays, that were preparedusing conventional synthetic routes. Selected reagentswere purchased from the following commercial sources:Endopeptidase Lys-C (Wako Chemicals, Osaka, Japan);porcine modified trypsin (Promega, Nepean, Ontario,Canada); loose ReproSil-Pur 120 C18-AQ 3 μm (DrMaisch, Ammerbuch-Entringen, Germany); 96-well fullskirt PCR plates (Axygen, Union City, CA, USA); fused sil-ica capillary tubing (Polymicro, Phoenix, AZ, USA); prote-ase inhibitor mixture (Roche Applied Science, Basel,Switzerland); NuPAGE Novex BisTris Gels (Invitrogen,Carlsbad, CA, USA). All cloning enzymes were from NewEngland Biolabs. Oligonucleotides were custom synthe-sized at Eurofins MWG GmbH, Ebersberg, Germany.Matrix for sample analysisThe isotopic labelling strategy employed here is limitedto triplexing so to enable a comparison of the proteinexpression profile for one colony to all others weemployed a design similar to what we have done previ-ously that maximized the statistical power of the experi-ment to detect effects in our parameters of interest [27].We collected three replicate samples from each colony,grouped the samples in blocks of three, assigned one ofthe three isotopic labels to each sample, and assignedcolonies to blocks in order to minimize the variance ofthe hygienic behavior variables. We constrained the ex-periment so no two colonies from the same populationwere in the same block and no two samples from thesame colony were assigned the same isotopic label. Thisensured the experimental design did not confound thehygienic behavior effect with the bee population or theisotopic label.Protein preparation for mass spectrometryBee antennae samples were washed three times withphosphate-buffered saline (PBS) and bead-homogenizedin buffer (50 mM Tris-Cl, 150 mM NaCl, 1% NP-40, 1%DTT) for three 20 s bursts at 6.5 M/s, with 1 min reston ice between each burst. Insoluble material was pel-leted at 600 relative centrifugal force (RCF) and proteinwas precipitated from the supernatants using 800 μL ofethanol, 20 μL of 2.5 M sodium acetate (pH 5.5) and2 μL of glycogen (10 mg/ml). The precipitation wasallowed to proceed at room temperature for 90 min.After centrifugation at 16,000 r.c.f. for 15 min, the pel-lets were dried and solubilized in buffered urea (6 Murea, 2 M thiourea, 100 mM Tris-Cl at pH 8.0, 20 mMDTT). Any insoluble material was then removed by cen-trifugation at 16,000 r.c.f. for 15 min. Protein concentra-tions were measured by a micro Bradford assay usingserial dilutions of bovine serum albumin to generate astandard curve. Protein samples were resolved on 1-DNu-PAGE (Invitrogen) gels and visualized with Coomas-sie Safe Blue (Pierce) to check the protein stability andquantity. For each sample, 20 μg of protein was dilutedGuarna et al. BMC Genomics  (2015) 16:63 Page 12 of 16to 1 μg/μl in urea buffer (6 M urea, 2 M thiourea,100 mM Tris-Cl, pH 8.0) before digestion [54].Peptide clean-up and labellingTen micrograms of digested peptides were purified withSTop And Go Extraction (STAGE) tips [55] and labelledvia reductive dimethylation using formaldehyde isotopo-logues [56,57]. In each triplex block one sample received10 μl of 200 mM CH2O (light) and 1 μl of 1 MNaBH3CN, one received 10 μl of 200 mM C2H2O(medium) and 1 μL of 1 M NaBH3CN and one received10 μL of 200 mM 13C2H2O (heavy) and 1 μL of 1 MNaBH3CN. The labelling reaction was performed twiceon each sample for 1 h each. The reactions were termi-nated by the addition of 20 μL of 3 M NH4Cl. Sampleswere adjusted to pH <3 by adding sample buffer (3% (w/v)acetonitrile, 1% (v/v) trifluoroacetic acid, 0.5% (v/v) aceticacid). Finally, 4 μg of each of the three differentially-labeled samples were combined and cleaned up again witha STAGE tip; two technical replicates were prepared foreach block, with one to be analyzed on the LTQ-FT andthe other on the LTQ-Orbitrap. Samples were stored onSTAGE tips at 4°C as needed.Liquid chromatography-tandem mass spectrometry(LC-MS/MS)Peptides were eluted from the STAGE tips using elu-tion buffer (0.1% trifluoroacetic acid, 80% acetonitrile).Then they were dried and resuspended in sample buf-fer (1% trifluoroacetic acid, 3% acetonitrile, 0.5% aceticacid). LC-MS/MS was performed using an 1100 Seriesnanoflow high performance liquid chromatographysystem (Agilent Technologies) on-line coupled to a lin-ear trapping quadrupole (LTQ)-Fourier transform (FT)or a LTQ-Orbitrap (ThermoFisher Scientific, Bremen,Germany). Peptide separation was performed by re-versed phase chromatography using a 75 μm innerdiameter fused silica emitter self-packed with 3 μmReprosil-Pur C18-AQ resin (Dr. Maisch). Peptides wereloaded in 4.8% (v/v) acetonitrile, 0.5% (v/v), acetic acidat 0.6 μL/min and then resolved at 200 nL/min for75 min using a linear gradient of acetonitrile from4.8% to 64% in 0.5% (v/v) acetic acid. Operating in datadependent mode, the LTQ-FT was set up to acquirefull scan data in the FT detector over a mass range of350–1600 m/z before performing FT selected ion mon-itoring (SIM) and MS/MS in the ion trap on the top 3most intense multiply charged ions [58]. The LTQ-OrbitrapXL was set to acquire a full-range scan at60,000 resolution from 350 to 1600 Th in the Orbitrapto simultaneously fragment the top five peptide ions ineach cycle in the LTQ (minimum intensity 1000 counts).Parent ions were then excluded from MS/MS for the next30 s. Singly-charged ions were excluded since in ESI modepeptides usually carry multiple charges. The Orbitrap wascontinuously recalibrated using the lock-mass function.Protein identification and quantificationFragment spectra peak lists were created using DTASu-perCharge [59] with default parameters. For each blockof samples, the peak list generated from LTQ-FT wascombined with the peak list from LTQ-Orbitrap beforeperforming Mascot search (v2.2) against the Honey Bee,A. mellifera Amel_4.0 translation (forward plus reversedsequences) of the genome with additional entries for hu-man keratins, porcine trypsin and LysC. Tryptic cleavagerules (R/K, except preceding a P) were specified with upto two missed cleavages allowed. Carbamidomethyl (C)was set as a fixed modification, Acetyl (Protein N-term),Deamidated (NQ), Oxidation (M), Dimethyl (K), Di-methyl (N-term), Dimethyl: 2H(4) (K), Dimethyl: 2H(4)(N-term), Dimethyl: 2H(6)13C(2) (K), Dimethyl: 2H(6)13C(2) (N-term) as variable modifications. Peptide toler-ance was set to 10 ppm and MS/MS tolerance was0.6 Da for the initial search. After recalibration of sys-tematic mass errors the peptide mass accuracy is typic-ally <2 ppm. The false discovery rate within each blockwas limited to 1%, estimated by counting the numberof ‘hits’ against the reversed sequences. Across thewhole experiment, however, the FDR approaches zerosince no reversed hits survived the filter requiring thata protein had to be detected in at least one quarter ofall blocks. All peptides with an IonsScore ≥25 were quan-tified using MSQuant (v1.5) [59]; after automated quanti-tation all files were manually edited to ensure consistentquantitation and the peak area ratios were exportedfor further analysis. An in-house script, finalList.pl(available here: http://www.chibi.ubc.ca/wp-content/up-loads/2013/01/finalList.pl_.txt) for applying parsimony(Occam’s razor) to generate a non-redundant list of identi-fied proteins from a large pool of independent experimentswas adapted to simultaneously calculate average peptideratios for each protein in each block. All raw data areavailable from the Honey Bee Peptide Atlas (http://www.peptideatlas.org/builds/honeybee/) and Proteo-meXchange (identifier PXD001616) while all the pro-teins identified and their quantitative ratios can befound in Additional file 4: Table S1.Statistical analysis and marker selectionIdentification of proteins whose expression patterns cor-related with population of behavioral data was per-formed as described previously [27]. Briefly, logarithmsof intensities were normalized by first subtracting theaverage of the three measurements in each block (foreach protein independently) and then centering andstandardizing within each label (across proteins) by themedian and median absolute deviation. For each protein,Guarna et al. BMC Genomics  (2015) 16:63 Page 13 of 16a Linear Mixed Effects model was used to estimate theeffect of each predictor variable, either population or hy-gienic behavior, on the protein expression level, adjust-ing for block and label factors. In the case of the BL-Y1dataset, analysis of the effect of hygienic behavior wasdone adjusting for population of origin. For the pre-dictor variables, an estimated effect, standard error andP-value were computed for each protein response. FDRs(Q-values) were computed for the set of P-values of agiven predictor over all protein response variables to ad-just for multiple comparisons. All calculations were per-formed in the R statistical language. In addition toselecting markers with low Q values from the BL-Y1dataset, we used the Y1 colonies from the two differentapiaries and selected proteins that had P < .05 across allfield parameters in both datasets. After completing theproteomic analysis of all Y1, Y2 and Y3 datasets, we fur-ther selected proteins by ordering them based on anoverall HB correlation factor The HB correlation wascomputed by adding a heritability factor to an average ofthe HB factors computed from each dataset. The HBfactor in each dataset was calculated by combining abiological and statistical factor as detailed in the Add-itional 5. The heritability factor for each protein, wasbased on a regression of the protein level observed inthe F1 daughters on the observed level in the paternal(Sir) and maternal (Dam) parent colonies (see Additional5 for more details).Expression clustering and gene ontology enrichmentSOTA (self-organizing tree algorithm) clustering wasused to determine one side probability metrics for allthirty-eight population-significant (Q ≤ .05) proteinsacross all honey bee populations. Using MultiExperimentViewer, six hard clusters were generated and hierarchicaldendrograms for population and proteins were con-structed using Euclidean distances [60]. For each cluster,gene ontology (GO) enrichment analysis was performedbased on the Drosophila orthologs to the complete pro-tein sequence of the bee proteins identified. DAVID(Database for Annotation, Visualization, and IntegratedDiscovery) [61,62] was used to calculate enrichments be-tween protein lists of interest using the entire identifiedantenna proteome characterized here (470 proteins) asbackground.Expression of recombinant proteins and binding assaysRNA extraction and cDNA synthesisTotal RNA was extracted using TRI® Reagent (Sigma),following the manufacturer’s protocol. cDNA was pre-pared from total RNA by reverse transcription, using 200units of SuperScript™ III Reverse Transcriptase (Invitro-gen) and 0.5 mg of an oligo-dT primer in a 50 μL reac-tion volume. The mixture also contained 0.5 mM ofeach dNTP (GE-Healthcare), 75 mM KCl, 3 mM MgCl2,10 mM DTT and 0.1 mg/ml BSA in 50 mM Tris–HCl,pH 8.3. The reaction mixture was incubated at 50°C for60 min and the product was directly used for PCR amplifi-cation or stored at −20°C.Polymerase chain reactionAliquots of 1 μL of crude cDNA were amplified in aBio-Rad Gene Cycler thermocycler, using 2.5 units ofThermus aquaticus DNA polymerase (GE-Healthcare),1 mM of each dNTP (GE-Healthcare), 1 μM of eachPCR primer, 50 mM KCl, 2.5 mM MgCl2 and 0.1 mg/mlBSA in 10 mM Tris–HCl, pH 8.3, containing 0.1% v/vTriton X-100. At the 5’ end, we used specific primerscorresponding to the sequence encoding the first fiveamino acids of the mature protein. The primers alsocontained an NdeI restriction site, for ligation into theexpression vector and providing at the same time theATG codon for an additional methionine in position 1.At the 3’ end specific primers were used, encoding thelast six amino acids, followed by a stop codon and anEcoRI restriction site for ligation into the expressionvector. Therefore, we used the following primers for theeach protein (enzyme restriction sites are italicized):fwAmelOBP16: 5’- GAGGAATAACATATGACACATGAGGAATT -3’rvAmelOBP16: 5’- GAATTCTTAGGAATTTAATATATCAGT -3’fwAmelOBP18: 5’- GAGGAATAACATATGACACTTGAAGAATT -3’rvAmelOBP18: 5’- GAATTCTTAGCCACTTAACATTTCTTT -3’After a first denaturation step at 95°C for 5 min, weperformed 35 amplification cycles (1 min at 95°C, 30 sat 50°C, 1 min at 72°C) followed by a final step of7 min at 72°C. We obtained amplification products of300–400 bp, in agreement with the expected sizes.Cloning and sequencingThe crude PCR products were ligated into a pGEM(Promega) vector without further purification, using a1:5 (plasmid:insert) molar ratio and incubating the mix-ture overnight, at room temperature. After transform-ation of E. coli XL-1 Blue competent cells with theligation products, positive colonies were selected by PCRusing the plasmid’s primers SP6 and T7 and grown inLB/ampicillin medium. DNA was extracted using thePlasmid MiniPrep Kit (Euroclone) and custom sequencedat Eurofins MWG (Martinsried, Germany).Cloning in expression vectorspGEM plasmids containing the appropriate sequenceswere digested with Nde I and Eco RI restriction enzymesfor 2 h at 37°C and the digestion products were separatedGuarna et al. BMC Genomics  (2015) 16:63 Page 14 of 16on agarose gel. The obtained fragments were purifiedfrom gel using QIAEX II Extraction kit (Qiagen) and li-gated into the expression vector pET5b (Novagen,Darmstadt, Germany), previously linearized with thesame enzymes. The resulting plasmids were sequencedand shown to encode the mature proteins.Preparation of the proteinsFor expression of recombinant proteins, each pET-5bvector containing the appropriate odorant-binding pro-tein (OBP) sequence was used to transform BL21(DE3)pLysS and BL21(DE3)Rosetta-gami E. coli cells, forOBP18 and OBP16 respectively. Protein expression wasinduced by addition of IPTG to a final concentration of0.4 mM when the culture had reached a value of O.D.600 = 0.8. Cells were grown for further 2 h at 37°C, inthe case of OBP18, while they were grown overnight at30°C for OBP16 expression. They were then harvestedby centrifugation and sonicated. After centrifugation,OBP16 was soluble while OBP18 was present as inclu-sion bodies. To solubilize it, the pellet from 1 L of cul-ture was dissolved in 10 mL of 8 M urea, 1 mM DTT in50 mM Tris buffer, pH 7.4, then diluted to 100 mL withTris buffer and dialysed three times against Tris buffer.Purification of the proteins was accomplished bycombinations of chromatographic steps anion-exchangeresins, such as DE-52 (Whatman), QFF or Mono-Q (GE-Healthcare), followed by gel filtration on Sephacryl-100 orSuperose-12 (GE-Healthcare) along with standard pro-tocols previously adopted for other odorant-bindingproteins [63,64]. The electrophoretic analysis of crudebacterial pellets and representative fractions from thelast purification steps for OBP16 and OBP18 are shownin Additional file 6: Figure S1.Fluorescence measurementsEmission fluorescence spectra were recorded on a JascoFP-750 instrument at 25°C in a right angle configuration,with a 1 cm light path quartz cuvette and 5 nm slits forboth excitation and emission. The protein was dissolvedin 50 mM Tris–HCl buffer, pH 7.4, while ligands wereadded as 1 mM stock solutions in methanol.Fluorescence binding assaysTo measure the affinity of the fluorescent ligand 1-NPN(N-phenyl-1-naphthylamine) to each protein, a 2 μM so-lution of the protein in 50 mM Tris–HCl, pH 7.4, was ti-trated with aliquots of 1 μM ligand in methanol to finalconcentrations of 2–16 μM. The probe was excited at337 nm and emission spectra were recorded between380 and 450 nm. The affinity of other ligands was mea-sured in competitive binding assays, using 1-NPN as thefluorescent reporter at 2 μM concentration and 2–16 μM concentrations of each competitor.To avoid the artefact provided by the strong fluores-cent signals observed in the presence of ligands capableof forming micelles, such as long-chain fatty acids, weused 0.2-0.6 μM concentrations of each competitor. Infact, when this happens, the probe can bind inside thehydrophobic core of the micelle, emitting a signal similarto that produced in the binding pocket of a protein.For determining binding constants, the intensity valuescorresponding to the maximum of fluorescence emissionwere plotted against free ligand concentrations. Boundligand was evaluated from the values of fluorescence in-tensity assuming that the protein was 100% active, witha stoichiometry of 1:1 protein:ligand at saturation. Thecurves were linearized using Scatchard plots. Dissoci-ation constants of the competitors were calculated fromthe corresponding IC50 values (concentrations of ligandshalving the initial fluorescence value of 1-NPN), usingthe equation: KD = [IC50]/1 + [1-NPN]/K1-NPN, [1-NPN]being the free concentration of 1-NPN and K1-NPN beingthe dissociation constant of the complex Protein/1-NPN.Availability of supporting dataAll mass spectrometry raw data used here is available ineither Peptide Atlas (http://www.peptideatlas.org/builds/honeybee/) or the ProteomeXchange Consortium [65]via the PRIDE partner repository with the dataset identi-fier PXD001616”.Additional filesAdditional file 1: Table S2. Clustering and functional enrichment data.The statistical analysis for the correlation of proteins and thegeographical origins of the colonies Protein annotations can be found intab 1. Tab 2 is the output of cluster analysis using the self organizingmaps (SOM) procedure. Tab 3 and 4 give the gene ontology terms foundenriched in protein clusters. Tab 5 provides a description of the columnheadings used in this supplementary data set.Additional file 2: Table S3. Protein-HB correlation data. Each of thefirst six tabs represents one generation at one breeding site. Protein ac-cession numbers and statistical values are listed. The last two tabs pro-vide protein heritability.Additional file 3: Table S4. Ligand binding data. Measured IC50 andcalculated dissociation constants (KD) for 29 ligands tested with threeOBPs of the honeybee. Values are all expressed in μM.Additional file 4: Table S1. Protein Expression and HB data. Relativeratios for all proteins found in at least 25% of the colonies in onebreeding site/generation. Each of the first six tabs represents onegeneration at one breeding site. Protein accession numbers are listed onthe left and block/colony combinations are listed across the top. The lasttab contains all HB data.Additional file 5: Biological and Statistical factor calculations. TheBiological factor is the magnitude of the effect scaled by the average standarddeviation from all the effects while the statistical factor is the negative log ofthe P-value scaled so a factor of 1.0 corresponds to a P-value of 0.01 calculatedas follows: Biological factor = ABS(value)/ Average(STDEV), where ABS: is theabsolute value, STDEV is the standard deviation and STDEV = SE*SQRT(DF),where SE is the standard error, SQRT(DF) is the square root of the degrees offreedom; Statistical factor =−LOG100(P-value). The use of LOG100 resulted inpositive numbers of the same magnitude as the ABS(values). This is equivalentdesigned and performed the statistical analyses. II, MMG, and PP expressedThis work was supported by funding from Genome Canada, Genome BritishMarketing Association, the University of British Columbia and Agri-Foodand larvae by African honey bee colonies (Apis mellifera scutellata).Guarna et al. BMC Genomics  (2015) 16:63 Page 15 of 16Canada’s Advancing Canadian Agriculture and Agri-Food (ACAAF) program.The latter is part of a Collective Outcome project with the cooperation ofthe Investment Agriculture Foundation of British Columbia, the AlbertaAgriculture and Food Council, the Manitoba Rural Adaptation Council, theOntario Agricultural Adaptation Council, the Conseil pour le développement del’agriculture du Québec, and Agri-Futures Nova Scotia. Mass spectrometryinfrastructure used in this project was supported by the Canada Foundation forInnovation, the British Columbia Knowledge Development Fund and the BritishColumbia Proteomics Network. LJF is the Canada Research Chair in QuantitativeProteomics. The funders had no role in study design, data collection andanalysis, decision to publish, or preparation of the manuscript.Author details1Department of Biochemistry & Molecular Biology, Centre forHigh-Throughput Biology, University of British Columbia, 2125 East Mall,Vancouver, BC V6T 1Z4, Canada. 2Beaverlodge Research Farm, Agriculture &Agri-Food Canada, Beaverlodge, AB T0H 0C0, Canada. 3Kettle Valley Queens,Grand Forks, BC, Canada. 4Department of Agriculture, Food and Environment,5Columbia, the British Columbia Honey Producers Association through theBoone-Hodgson-Wilkinson Fund, the Canadian Honey Council and CanadianAssociation of Professional Apiculturists through the Canadian Bee ResearchFund, the British Columbia Blueberry Council, the British Columbia Cranberrythe recombinant OBPs and designed and performed the ligand bindingexperiments. MMG, KMM, QWTC and LJF interpreted the findings. MMG, IIand LJF wrote the first draft of the manuscript. All authors read andapproved the final manuscript.AcknowledgementsThe authors wish to thank Susan Cobey for her assistance with instrumentalinsemination, Asheber Sewalem for his advice in the structure of the diallelcrosses, members of our respective groups for continual support, and MartinAloise, Julian Yiu and Erin Boyle for assistance collecting and processing fieldsamples for Mass Spectrometry.as defining a unit of the statistical factor as the negative log of 0.01 with base100, −LOG100 of 0.01 = 1. A heritability value for each protein was estimatedby fitting a regression model that predicted the level observed in the F1daughter by the level observed in the paternal (Sir) and maternal (Dam) parentcolonies. This gave us estimated effects for the protein levels of Sir and Damparents as they relate to predicting F1 daughter protein levels. We thenassigned a value based on whether the effect between the F1 progeny andthe parents was higher than the median. If this was true for the two F0 parentsa value of 2 was assigned. If it was true for only one F0 parent or for none ofthem, a value of 1 or 0 was assigned, respectively.Additional file 6: Figure S1. OBP purification. SDS-PAGE analysis ofselected fractions from the last purification step of recombinant OBP16 andOBP18. M: molecular weight markers: 66, 45, 29, 24, 20 and 14 kDa. ni:bacterial pellet before induction with IPTG, i: bacterial pellet after induction.QFF: fractions containing purified OBP16 as eluted from strong anionexchange column QFF (GE-healthcare). S-100: fractions containing purifiedOBP18 as eluted from gel filtration column Sepharose-100 (Ge-Healthcare).Competing interestsEH is co-owner of Kettle Valley Queens, a company focused on breeding andselling honey bee queens.Authors’ contributionsLJF and SFP conceived of the initial idea for the study. MMG, APM, RW, SFP,LJF and EH designed the breeding experiments. APM, EH and SFP bred thebees, collected samples and measured hygienic behavior. RP, AT, KMM,MMG, NS and LJF collected antennal samples and performed the massspectrometry and analyzed mass spectrometry data. RW, LJF and MMGUniversity of Pisa, Pisa, Italy. Department of Statistics, University of BritishColumbia, Vancouver, BC V6T 1Z4, Canada. 6Current address: DalhousieUniversity, Halifax, NS, Canada. 7Current address: Macquarie University,Sydney, NSW, Australia.Apidologie. 2004;35:31–6.24. Ibrahim A, Spivak MS. The relationship between hygienic behavior andsuppression of mite reproduction as honey bee (Apis mellifera) mechanismsReceived: 23 May 2014 Accepted: 22 December 2014References1. Calderone NW. Insect pollinated crops, insect pollinators and US agriculture:trend analysis of aggregate data for the period 1992–2009. PLoS One.2012;7(5):e37235.2. van der Zee R, Brodschneider R, Brusbardis V, Charrière JD, Chlebo R, CoffeyMF, et al. Results of international standardised beekeeper surveys of colonylosses for winter 2012–2013: analysis of winter loss rates and mixed effectsmodelling of risk factors for winter loss. J Apicultural Res. 2014;53:19–34.3. United States Department of Agriculture. Report on the nationalstakeholders conference on honey bee health. 2012.4. Genersch E. American foulbrood in honeybees and its causative agent.Paenibacillus larvae. J Invertebr Pathol. 2010;103 Suppl 1:S10–9.5. Rosenkranz P, Aumeier P, Ziegelmann B. Biology and control of Varroadestructor. J Invertebr Pathol. 2010;103 Suppl 1:S96–119.6. Thompson TS, Noot DK, Calvert J, Pernal SF. Determination of lincomycinand tylosin residues in honey by liquid chromatography/tandem massspectrometry. Rapid Commun Mass Spectrom. 2005;19(3):309–16.7. Evans JD. Diverse origins of tetracycline resistance in the honey beebacterial pathogen Paenibacillus larvae. J Invertebr Pathol. 2003;83(1):46–50.8. Martin SJ, Elzen PJ, Rubink WR. Effect of acaricide resistance on reproductiveability of the honey bee mite Varroa destructor. Exp Appl Acarol.2002;27(3):195–207.9. Pettis JS. A scientific note on Varroa destructor resistance to coumaphos inthe United States. Apidologie. 2004;35(1):91–2.10. Elzen PJ, Westervelt D. Detection of coumaphos resistance in Varroadestructor in Florida. Am Bee J. 2002;142(4):291–2.11. Elzen PJ, Eischen FA, Baxter JR, Elzen GW, Wilson WT. Detection ofresistance in US Varroa jacobsoni oud. (mesostigmata : Varroidae) to theacaricide fluvalinate. Apidologie. 1999;30(1):13–7.12. Huang WF, Solter LF, Yau PM, Imai BS. Nosema ceranae escapes fumagillincontrol in honey bees. PLoS Pathog. 2013;9(3):e1003185.13. Elzen PJ, Baxter JR, Spivak M, Wilson WT. Control of Varroa jacobsoni oud.resistant to fluvalinate and amitraz using coumaphos. Apidologie.2000;31:437–41.14. Spivak M, Gilliam M. Hygienic behaviour of honey bees and its applicationfor control of brood diseases and Varroa. part I. Hygienic behaviour andresistance to American foulbrood. Bee World. 1998;79:124–34.15. Woodrow AW, Holst EC. The mechanism of colony resistance to Americanfoulbrood. J Econ Entomol. 1942;35:327–30.16. Spivak MS, Gilliam M. Facultative expression of hygienic behaviour of honeybees in relation to disease resistance. J Apicultural Res. 1993;32(3):147–57.17. Harbo J, Harris J. Suppressed mite reproduction explained by the behaviourof adult bees. J Apicultural Res. 2005;44(1):21–3.18. Currie RW, Gatien P. Timing acaricide treatments to prevent Varroadestructor (acari: Varroidae) from causing economic damage to honey beecolonies. Can Entomologist. 2006;138(2):238–52.19. Arechavaleta-Velasco ME, Alcala-Escamilla K, Robles-Rios C, Tsuruda JM, HuntGJ. Fine-scale linkage mapping reveals a small set of candidate genesinfluencing honey bee grooming behavior in response to Varroa mites.PLoS One. 2012;7(11):e47269.20. Spivak MS, Reuter GS. Resistance to American fouldbrood disease by honeybee colonies, Apis mellifera, bred for hygienic behavior. Apidologie.2001;32:555–65.21. Spivak MS, Gilliam M. Hygienic behavior of honey bees and its applicationfor control of brood diseases and varroa; part II. Studies on hygienicbehavior since the rothenbuhler era. Bee World. 1998;79:169–86.22. Villegas AJ, Villa JD. Uncapping of pupal cells by European bees in theUnited States as responses to Varroa destructor and Galleria mellonella.J Apicultural Res. 2006;45:203–6.23. Neumann P, Härtel S. Removal of small hive beetle (Aethina tumida) eggsof resistance to Varroa destructor. Apidologie. 2006;37:31–40.25. Parker R, Guarna MM, Melathopoulos AP, Moon KM, White R, Huxter E, et al.Correlation of proteome-wide changes with social immunity behaviors1987;18:27–42.54. Foster LJ, De Hoog CL, Mann M. Unbiased quantitative proteomics of lipidrafts reveals high specificity for signaling factors. Proc Natl Acad Sci U S A.2003;100(10):5813–8.55. Ishihama Y, Rappsilber J, Mann M. Modular stop and go extraction tips withstacked disks for parallel and multidimensional peptide fractionation inproteomics. J Proteome Res. 2006;5(4):988–94.56. Chan QW, Foster LJ. Changes in protein expression during honey bee larvalGuarna et al. BMC Genomics  (2015) 16:63 Page 16 of 16provides insight into resistance to the parasitic mite, Varroa destructor, inthe honey bee (Apis mellifera). Genome Biol. 2012;13(9):R81-2012-13-9-r81.26. Le Conte Y, Alaux C, Martin JF, Harbo JR, Harris JW, Dantec C. Socialimmunity in honeybees (apis mellifera): transcriptome analysis ofVarroa-hygienic behaviour. Insect Mol Biol. 2011;20(3):399–408.27. Parker R, Melathopoulos AP, White R, Pernal SF, Guarna MM, Foster LJ.Ecological adaptation of diverse honey bee (Apis mellifera) populations.PLoS One. 2010;5(6):e11096.28. Foret S, Maleszka R. Function and evolution of a gene family encodingodorant binding-like proteins in a social insect, the honey bee (Apis mellifera).Genome Res. 2006;16(11):1404–13.29. Iovinella I, Dani FR, Niccolini A, Sagona S, Michelucci E, Gazzano A, et al.Differential expression of odorant-binding proteins in the mandibular glandsof the honey bee according to caste and age. J Proteome Res.2011;10(8):3439–49.30. Swanson JA, Torto B, Kells SA, Mesce KA, Tumlinson JH, Spivak M. Odorantsthat induce hygienic behavior in honeybees: Identification of volatilecompounds in chalkbrood-infected honeybee larvae. J Chem Ecol.2009;35(9):1108–16.31. Schoning C, Gisder S, Geiselhardt S, Kretschmann I, Bienefeld K, Hilker M,et al. Evidence for damage-dependent hygienic behaviour towards Varroadestructor-parasitised brood in the western honey bee, Apis mellifera. J ExpBiol. 2012;215(Pt 2):264–71.32. Pelosi P, Zhou JJ, Ban LP, Calvello M. Soluble proteins in insect chemicalcommunication. Cell Mol Life Sci. 2006;63(14):1658–76.33. Yao M, Rosenfeld J, Attridge S, Sidhu S, Aksenov V, Rollo CD. The ancientchemistry of avoiding risks of predation and disease. Evolutionary Biology.2009;36:267–8.34. Garrido PM, Antunez K, Martin M, Porrini MP, Zunino P, Eguaras MJ.Immune-related gene expression in nurse honey bees (Apis mellifera)exposed to synthetic acaricides. J Insect Physiol. 2013;59(1):113–9.35. Pernal SF, Sewalem A, Melathopoulos AP. Breeding for hygienic behaviourin honeybees (Apis mellifera) using free-mated nucleus colonies. Apidologie.2012;43:403–6.36. Lapidge KL, Oldroyd BP, Spivak M. Seven suggestive quantitative trait lociinfluence hygienic behavior of honey bees. Naturwissenschaften.2002;89(12):565–8.37. Tsuruda JM, Harris JW, Bourgeois L, Danka RG, Hunt GJ. High-resolutionlinkage analyses to identify genes that influence Varroa sensitive hygienebehavior in honey bees. PLoS One. 2012;7(11):e48276.38. Wilson-Rich N, Spivak M, Fefferman NH, Starks PT. Genetic, individual, andgroup facilitation of disease resistance in insect societies. Annu RevEntomol. 2009;54:405–23.39. Leal WS. Odorant reception in insects: Roles of receptors, binding proteins,and degrading enzymes. Annu Rev Entomol. 2013;58:373–91.40. Tegoni M, Campanacci V, Cambillau C. Structural aspects of sexual attractionand chemical communication in insects. Trends Biochem Sci.2004;29(5):257–64.41. Chan QW, Chan MY, Logan M, Fang Y, Higo H, Foster LJ. Honey bee proteinatlas at organ-level resolution. Genome Res. 2013;23(11):1951–60.42. Dani FR, Iovinella I, Felicioli A, Niccolini A, Cavello MA, Carucci MG, et al.Mapping the expression of soluble olfactory proteins in the honeybee.J Proteome Res. 2010;9(4):1822–33.43. Keeling CI, Plettner E, Slessor KN. Hymenopteran semiochemicals. Top CurrChem. 2004;239:133–77.44. Schiavo G, Benfenati F, Poulain B, Rossetto O, Polverino De Laureto P,DasGupta BR, et al. Tetanus and botulinum-B neurotoxins blockneurotransmitter release by proteolytic cleavage of synaptobrevin. Nature.1992;359(6398):832–5.45. Whitfield CW, Cziko AM, Robinson GE. Gene expression profiles in the brainpredict behavior in individual honey bees. Science. 2003;302(5643):296–9.46. Marr HS, Edgell CJ. Testican-1 inhibits attachment of neuro-2a cells. MatrixBiol. 2003;22(3):259–66.47. Santelli E, Leone M, Li C, Fukushima T, Preece NE, Olson AJ, et al. Structuralanalysis of Siah1-siah-interacting protein interactions and insights into theassembly of an E3 ligase multiprotein complex. J Biol Chem. 2005;280(40):34278–87.48. Sumegi M, Hunyadi-Gulyas E, Medzihradszky KF, Udvardy A. 26S proteasomesubunits are O-linked N-acetylglucosamine-modified in Drosophilamelanogaster. Biochem Biophys Res Commun. 2003;312(4):1284–9.development. Genome Biol. 2008;9(10):R156.57. Boersema PJ, Aye TT, Van Veen TA, Heck AJ, Mohammed S. Triplex proteinquantification based on stable isotope labeling by peptide dimethylationapplied to cell and tissue lysates. Proteomics. 2008;8(22):4624–32.58. Chan QW, Howes CG, Foster LJ. Quantitative comparison of caste differencesin honeybee hemolymph. Mol Cell Proteomics. 2006;5(12):2252–62.59. Mortensen P, Gouw JW, Olsen JV, Ong SE, Rigbolt KTG, Bunkenborg J, et al.MSQuant, an open source platform for mass spectrometry-basedquantitative proteomics. J Proteome Res. 2010;9:393–403.60. Saeed AI, Sharov V, White J, Li J, Liang W, Bhagabati N, et al. TM4: a free,open-source system for microarray data management and analysis.Biotechniques. 2003;34(2):374–8.61. Dennis Jr G, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, et al. DAVID:database for annotation, visualization, and integrated discovery. GenomeBiol. 2003;4(5):3.62. Da Huang W, Sherman BT, Lempicki RA. Systematic and integrative analysisof large gene lists using DAVID bioinformatics resources. Nat Protoc.2009;4(1):44–57.63. Ban L, Scaloni A, Brandazza A, Angeli S, Zhang L, Yan Y, et al. Chemosensoryproteins of Locusta migratoria. Insect Mol Biol. 2003;12(2):125–34.64. Calvello M, Guerra N, Brandazza A, D’Ambrosio C, Scaloni A, Dani FR, et al.Soluble proteins of chemical communication in the social wasp Polistesdominulus. Cell Mol Life Sci. 2003;60(9):1933–43.65. Vizcaíno JA, Deutsch EW, Wang R, Csordas A, Reisinger F, Ríos D, et al.ProteomeXchange provides globally co-ordinated proteomics datasubmission and dissemination. Nature Biotechnol. 2014;30(3):223–6.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistribution49. Satoh T, Ishizuka T, Tomaru T, Yoshino S, Nakajima Y, Hashimoto K, et al.Tat-binding protein-1 (TBP-1), an ATPase of 19S regulatory particles of the26S proteasome, enhances androgen receptor function in cooperation withTBP-1-interacting protein/Hop2. Endocrinology. 2009;150(7):3283–90.50. Spivak M, Reuter GS. Varroa destructor infestation in untreated honey bee(hymenoptera: Apidae) colonies selected for hygienic behavior. J EconEntomol. 2001;94(2):326–31.51. Rinderer TE, Harris JW, Hunt GJ, De Guzman LI. Breeding for resistance toVarroa destructor in North America. Apidologie. 2010;41(3):409–24.52. Harbo JR, Harris JW. Responses to Varroa by honey bees with differentlevels of Varroa sensitive hygiene. J Apic Res. 2009;48(3):156–61.53. Moritz RFA, Southwick EE, Harbo JR. Genetic analysis of defensive behaviourof honeybee colonies (Apis mellifera L.) in a field test. Apidologie.Submit your manuscript at www.biomedcentral.com/submit


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items